Cauchy noise removal using group-based low-rank prior

Abstract Although the extensive research on Gaussian noise removal, few works consider the Cauchy noise removal problem. In this paper, we propose a novel group-based low-rank method for Cauchy noise removal. By exploiting the nonlocal self-similarity of natural images, we consider a group of similar patches as an approximate low-rank matrix, and formulate the denoising of each group as a low-rank matrix recovery problem. Meanwhile, we develop the alternating direction method of multipliers algorithm to solve the proposed nonconvex model with guaranteed convergence. Experiments illustrate that our method has superior performance over the state-of-the-art methods in terms of both visual and quantitative measures.